Evaluating the Cardiovascular Safety of New Medications for Type 2 Diabetes: Time to Reassess?
The U.S. Food and Drug Administration (FDA) issued a Guidance for Industry in 2008 defining preapproval and postapproval requirements for the demonstration of cardiovascular safety for all new medications developed for glycemic management in type 2 diabetes. Seventeen large, prospective, randomized, controlled clinical trials involving more than 140,000 subjects thus far have been completed or are ongoing in accordance with this guidance. All five of the completed trials, involving three different drug classes, have met their primary objective to exclude an unacceptable level of ischemic cardiovascular risk as defined in the FDA guidance. Additionally, one trial found an increased risk of hospitalization for heart failure, and another demonstrated decreases in cardiovascular mortality and hospitalization for heart failure. Given that a heightened risk of cardiovascular ischemic events has not been demonstrated across several classes of new diabetes drugs, we believe it is time for the scientific community and the FDA to consider a more targeted approach to what is, in effect, a global cardiovascular safety trial requirement for all new type 2 diabetes medications in development.
Multiple new medications for glycemic management of type 2 diabetes have recently been approved by the U.S. Food and Drug Administration (FDA), with additional drugs in development. In the regulatory process, all agents for treatment of hyperglycemia in diabetes are evaluated for efficacy in lowering blood glucose levels and overall safety and are subjected to particularly rigorous screening for cardiovascular safety. The approach to these cardiovascular studies is largely directed by a Guidance for Industry issued by the FDA in 2008 stating that all new type 2 diabetes drug development programs should rule out unacceptable cardiovascular risk by demonstrating an upper bound of the two-sided 95% CI of the risk ratio <1.8 preapproval for a composite end point of major adverse cardiac events (MACE), consisting of at least cardiovascular death, nonfatal myocardial infarction, and stroke (1). For drugs in which the upper bound of the CI is between 1.3 and 1.8, the guidance states that subsequent postapproval trials of the same composite MACE end point should be conducted to continue marketing. These recommendations were motivated by the high prevalence of cardiovascular disease in diabetes (accounting for approximately 70% of deaths) and, in part, by concerns about a signal of potentially increased cardiovascular risk for the thiazolidinedione rosiglitazone (2) occurring on a background of additional cardiac safety concerns with antihyperglycemic medications for type 2 diabetes. The latter included evidence for increased deaths and major cardiovascular events during the development program of the peroxisome proliferator–activated receptor (PPAR) agonist muraglitazar (3), persistent uncertainty about cardiovascular safety with sulfonylureas (4), increased mortality with intense glucose control in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial (5), and increased risk for congestive heart failure with pioglitazone (6) and rosiglitazone (7). Seventeen large, randomized, controlled, multicenter postapproval trials involving more than 140,000 participants thus far have been initiated in accordance with the FDA guidance (Tables 1 and 2). Results from five of these trials now have been reported, providing an opportunity to examine the utility of this approach to evaluate the safety of new antihyperglycemic medications for type 2 diabetes.
The FDA Endocrinologic and Metabolic Drugs Advisory Committee was convened in April 2015 to review the results of postmarketing trials of the dipeptidyl peptidase 4 (DPP-4) inhibitors saxagliptin (Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus–Thrombolysis in Myocardial Infarction 53 [SAVOR-TIMI 53] trial) and alogliptin (Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care [EXAMINE] trial) in type 2 diabetes. The study data reviewed at those meetings demonstrated hazard ratios for a composite MACE end point (cardiovascular death, nonfatal myocardial infarction, and nonfatal stroke) of 1.00 for saxagliptin and 0.96 for alogliptin. The upper bound of the 95% CI for the two drugs was 1.12 and 1.16, respectively, which was below a threshold for concern based on the FDA guidance (Table 1) (8,9). The advisory committee vote on whether acceptable cardiovascular risk profiles were demonstrated was 13 “yes,” 1 “no,” and 1 abstention for saxagliptin and was unanimous at 16 “yes” for alogliptin (note that there was not complete overlap in members attending each meeting). Results of three additional cardiovascular outcome trials conducted under the FDA guidance subsequently have been published on the DPP-4 inhibitor sitagliptin (Trial Evaluating Cardiovascular Outcomes with Sitagliptin [TECOS] trial) (10), the sodium–glucose cotransporter 2 (SGLT2) inhibitor empagliflozin (BI 10773 [Empagliflozin] Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients [EMPA-REG OUTCOME] study) (11), and the glucagon-like peptide 1 (GLP-1) receptor agonist lixisenatide (Evaluation of Lixisenatide in Acute Coronary Syndrome [ELIXA] study) (12). The sitagliptin and lixisenatide studies demonstrated a hazard ratio for a composite MACE end point approximating unity and an upper bound of the 95% CI <1.30 for both drugs, while the empagliflozin results provided evidence for cardiovascular benefit with a hazard ratio of 0.86 and upper bound of the 95% CI 0.99 (Table 1). Thus, all five trials involving medications from three different drug classes have excluded an unacceptable level of ischemic cardiovascular risk (which was the primary analysis) as defined in the FDA 2008 guidance document.
All-cause mortality, another important, prespecified secondary end point, raised focused concerns by demonstrating inconsistent findings across the DPP-4 inhibitor trials. For saxagliptin, all-cause mortality in the intent-to-treat analysis population showed a hazard ratio of 1.11 (95% CI 0.96–1.27), which increased to 1.23 (95% CI 1.02–1.48) when evaluated in the on-treatment population. It is important to note that, consistent with the lack of increased risk of MACE, the increased mortality risk was primarily driven by noncardiovascular deaths, with no specific identified causal concerns. By contrast, all-cause mortality was not numerically increased with alogliptin, sitagliptin, or lixisenatide, and all-cause mortality was significantly decreased with empagliflozin (Table 1). It is not possible to directly compare agents as these trials had different inclusion criteria and participants were not randomized across the various medications.
In addition to concerns about ischemic cardiovascular events, as addressed by the recommendations for a composite MACE in the FDA guidance, there is an increased risk of heart failure in patients with diabetes (13). Increased heart failure risk has previously been demonstrated with thiazolidinediones (6,7) but was not apparent during the initial development of any of the agents for which postmarketing safety trials were subsequently performed under the 2008 FDA guidance. Unexpectedly, exploratory analysis of hospitalization for heart failure as an individual clinical end point revealed an elevated hazard ratio for saxagliptin at 1.27 (95% CI 1.07–1.51) and a nonsignificant numerical increase in risk for alogliptin with a hazard ratio of 1.19 (95% CI 0.90–1.58). The hazard ratios for hospitalization for heart failure as a prespecified, adjudicated individual end point approximated unity in the sitagliptin and lixisenatide trials (Table 1), and increased risk for an expanded MACE end point that included hospitalization for heart failure was not evident with any of these incretin-modulating drugs (10,12). By contrast, the empagliflozin trial showed a decreased hazard ratio for hospitalization for heart failure as an individual end point at 0.65 (95% CI 0.50–0.85) and as a component of an expanded MACE composite outcome combined with CV death but excluding stroke at 0.66 (95% CI 0.55–0.79) (11). Thus, hospitalization for heart failure as a stand-alone exploratory end point demonstrated inconsistent results for the DPP-4 inhibitors, with a potential concern observed for saxagliptin, a less evident trend for alogliptin, and no numeric or statistical difference for sitagliptin. Hospitalization for heart failure risk was either neutral or decreased with the other two agents studied. On the basis of the findings in the SAVOR-TIMI 53 and EXAMINE trials, the FDA Endocrinologic and Metabolic Drugs Advisory Committee recommended a label warning regarding the risk of heart failure and further heart failure safety monitoring for saxagliptin and alogliptin. It is important to note that the evidence for increased heart failure risk with these two pharmacological agents requires confirmation. Additionally, as none of the recently reported trials conducted under the 2008 FDA guidance focused on a population with a high risk that would generate an adequate event rate for heart failure, the capacity of these completed studies to detect heart failure risk with certainty may be limited.
All of these cardiovascular outcome trials (including the empagliflozin study) were primarily designed to exclude unacceptable risk of cardiovascular ischemic events by first testing for noninferiority relative to placebo on the primary MACE or expanded MACE end point in accordance with the FDA guidance. Additionally, the analysis plans permitted secondary testing for superiority over placebo if the noninferiority threshold was achieved, providing an opportunity to show potential cardiovascular benefit. In the context of the trial designs, which included mean follow-up duration ≤3 years and modest or no glycemic differences in the placebo and usual care groups, cardiovascular benefit was only demonstrated for the SGLT2 inhibitor empagliflozin (11). This represents the first large-scale, prospective, randomized, controlled cardiovascular outcome trial of any antihyperglycemic medication for type 2 diabetes showing cardiovascular benefit, although smaller studies have seen benefit with other agents. The mechanism and consistency of the benefit across the SGLT2 inhibitor class remains to be defined and cannot be fully established based on the results of a single trial where the primary intent was to demonstrate exclusion of cardiovascular risk. A recent press release for the LEADER (Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results) trial investigating the cardiovascular safety of the GLP-1 receptor agonist liraglutide reports reduced cardiovascular risk assessed by the composite outcome of the first occurrence of cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke (14). However, the data from this study have not yet been made available. The 2008 FDA guidance defines a need for cardiovascular safety evaluation but not for assessment of cardiovascular outcome benefit, and the FDA does not otherwise require new diabetes drugs with the intended indication of improving glycemic control to also show cardiovascular benefit. The current FDA guidance to exclude cardiovascular risk of new diabetes medications therefore leads to a trial design that focuses on the risk of the drug per se and allows for use and dose adjustments of multiple diabetes drugs other than the specific agent under study. The cardiovascular outcome studies occur on a backdrop of established drug efficacy for glycemic control, and these safety trials thus are not designed to examine either glycemic efficacy or glycemic effects as potential components of drug superiority in lowering cardiovascular risk. Furthermore, to accrue sufficient events in a timely manner, these trials generally enroll patients with more advanced atherosclerotic cardiovascular risk or established cardiovascular disease. The typical trial observation periods are most effective for assessing short-term cardiovascular risk but do not permit assessment of long-term risks or benefits for these comorbid conditions.
In reviewing the saxagliptin and alogliptin trials, multiple FDA advisory committee members questioned the wisdom of continuing to impose the burden of cardiovascular outcome safety trials on all new diabetes medications in order to achieve full market approval. They noted that concern about the potential for excessive cardiovascular risk with diabetes therapies has been somewhat mitigated by the readjudication of the original signal from the rosiglitazone cardiovascular outcome trial, which demonstrated that there was no imbalance in MACE events despite earlier doubts raised from meta-analyses of the phase 2 trial data (2). It is now apparent that the concern about increased ischemic cardiovascular disease risk with diabetes medications, which has been broadly applied by the FDA to all new antihyperglycemic type 2 diabetes drugs in development regardless of drug class or prior signals of cardiovascular risk, has not been substantiated in large-scale trials on three DPP-4 inhibitors (saxagliptin, alogliptin, and sitagliptin), the GLP-1 receptor agonist lixisenatide, and the SGLT2 inhibitor empagliflozin. This new information should be incorporated into the totality of evidence to assess the impact and unintended consequences of the 2008 FDA guidance being applied broadly to diabetes drug development.
The history leading up to the FDA 2008 guidance, combined with the new evidence on the DPP-4 inhibitor, GLP-1 receptor agonist, and SGLT2 inhibitor drug classes, raises the issue of how to best assess all aspects of cardiac safety in the development of new diabetes medications. The additional 12 ongoing cardiovascular outcome studies include three with DPP-4 inhibitors, six with GLP-1 receptor agonists, and three with SGLT2 inhibitors and in total involve more than 90,000 additional patients (Table 2). The results of these trials will undoubtedly expand the knowledge of the cardiovascular and heart failure risks of new diabetes drugs and will likely reveal a better understanding of the potential additional safety concerns. However, we question whether continued broad application of the current FDA guidance on cardiovascular safety to all new diabetes drugs, including additional members of already studied drug classes, is justified. Each new large-scale cardiovascular outcome trial is accompanied by high financial costs that add directly or indirectly to the overall health care budget in the U.S. and potentially divert funding away from new research efforts.
The argument to continue the broad requirement to perform large cardiovascular outcome trials for all new diabetes drugs to simply extend the patient-years of exposure beyond the phase 3 trial requirement for approval to identify all unforeseen safety concerns could be made for any drug class. In the case of diabetes drugs, the AleCardio trial was a study enrolling 7,226 patients with a recent acute coronary syndrome event to determine if the dual activator of PPAR-α and PPAR-γ aleglitazar would reduce the risk of cardiovascular events (15). The trial was stopped early due to futility (lack of superiority), and, at that time, several new safety concerns also were identified. The latter included adverse events previously associated with other PPAR-α or PPAR-γ activators (heart failure, decreased glomerular filtration, and bone fractures), plus evidence for increased risk of gastrointestinal bleeding, which was unanticipated. Of note, there was a numerical increase in the risk of hospitalization from heart failure with a hazard ratio of 1.22, but this was not confirmed to be significant at the time the study was terminated (95% CI 0.94–1.59) and thus remains unresolved. The identification of gastrointestinal bleeding in the AleCardio study as a risk not previously associated with agents sharing a similar mechanism of action is of interest from a general safety consideration. However, the current FDA recommendation to perform cardiovascular safety trials for all new diabetes drugs is specifically intended to identify specific cardiovascular risks, not all unanticipated safety concerns. The FDA has a variety of postmarketing surveillance mechanisms to identify safety concerns that are appropriate in the context of any drug requiring broad safety oversight.
The 2008 FDA guidance on the evaluation of cardiovascular safety in the development of new antihyperglycemic medications for type 2 diabetes was thoughtfully designed based on the evidence available at that time. However, we believe the scientific community and the FDA now should reconsider the totality of the current evidence and engage in a discussion on safety assessment and monitoring strategies that would represent a more targeted approach to cardiovascular safety trials for new diabetes medications. We support the overall FDA process for evaluating new antihyperglycemic drugs, which requires adequate data to demonstrate efficacy in lowering blood glucose and to provide reasonable assurance of overall safety prior to initial drug approval. For medications that achieve approval, the FDA is empowered to mandate additional targeted postapproval safety studies to address concerning specific safety signals that are identified either during drug development or postmarketing. This process has been effective in identifying and evaluating cardiovascular safety concerns and also other safety issues, as exemplified by the recent required investigation of potential increased risks of pancreatitis and pancreatic cancer with DPP-4 inhibitors and GLP-1 receptor agonists. We question whether a continued recommendation for large-scale cardiovascular safety trials to rule out a risk of ischemic events for all new antihyperglycemic medications, as detailed in the FDA 2008 Guidance for Industry, is cost-effective and justified. We believe it would be more effective to conduct such studies only when concerning safety data in the preapproval package, plausible mechanisms of risk, or a known class effect support a large investment in a cardiovascular outcome trial. If the preapproval package or postapproval monitoring provides a signal of other cardiac risks, such as heart failure, then trials specifically designed to evaluate those risks should be performed. Thus, we support a more informed approach to safety trials that would direct a more individualized strategy for addressing concerns about the cardiac safety of new diabetes medications.
Duality of Interest. R.J.S. has served on the FDA Endocrinologic and Metabolic Drugs Advisory Committee since 2011, has been chair of the committee since 2014, and chaired the 14 April 2015 advisory meeting on the SAVOR-TIMI 53 and EXAMINE trials. A.B.G. served on the FDA Endocrinologic and Metabolic Drugs Advisory Committee from 2007 to 2012 and previously served on a scientific advisory board for Novo Nordisk with no ongoing obligation. She provided an invited, independent commentary on the TECOS trial at the 2015 American Diabetes Association Scientific Sessions. W.R.H. has served on FDA advisory committees since 2003, has been an active member of the FDA Endocrinologic and Metabolic Drugs Advisory Committee since 2013, and participated in the 14 April 2015 advisory meeting on the SAVOR-TIMI 53 and EXAMINE trials. No other potential conflicts of interest relevant to this article were reported.
- Received October 13, 2015.
- Accepted February 23, 2016.
- © 2016 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.